IJE TRANSACTIONS B: Applications Vol. 29, No. 11, (November 2016) 1499-1506 Please cite this article as: I. Ebtehaj, H.Bonakdari, A Comparative Study of Extreme Learning Machines and Support Vector Machines in Prediction of Sediment Transport in Open Channels, International Journal of Engineering (IJE), TRANSACTIONS B: Applications Vol. 29, No. 11, (November 2016) 1499-1506 International Journal of Engineering Journal Homepage: www.ije.ir A Comparative Study of Extreme Learning Machines and Support Vector Machines in Prediction of Sediment Transport in Open Channels I. Ebtehaj a,b , H.Bonakdari *a,b a Department of Civil Engineering, Razi University, Kermanshah, Iran b Water and Wastewater Research Center, Razi University, Kermanshah, Iran PAPER INFO Paper history: Received 10 August 2016 Received in revised form 27 September 2016 Accepted 30 September 2016 Keywords: Extreme Learning Machines (ELM) Non-deposition Open channel Sediment transport Support Vector Machines (SVM) A B S T RA C T The limiting velocity in open channels to prevent long-term sedimentation is predicted in this paper using a powerful soft computing technique known as Extreme Learning Machines (ELM). The ELM is a single Layer Feed-forward Neural Network (SLFNN) with a high level of training speed. The dimensionless parameter of limiting velocity which is known as the densimetric Froude number ( Fr) is predicted using ELM and the results are compared to those obtained using a Support Vector Machines (SVM). The comparison of the ELM and SVM methods indicates a good performance for both methods in the prediction of Fr. In addition to being computationally faster, the ELM method has a higher level of accuracy (R 2 =0.99, MAE=0.10; MAPE=2.34; RMSE=0.14; CRM=0.02) compared with the SVM approach. doi: 10.5829/idosi.ije.2016.29.11b.03 NOMENCLATURE A cross-sectional area of flow (m/s 2 ) s specific gravity of sediment b bias terms of the equation V flow velocity (m/s) bi threshold of the i th hidden neuron w weighting vector CV volumetric sediment concentration y flow depth (m) D pipe diameter (m) Greek Symbols d median particle diameter (m) λs sediment friction factor Dgr (=((d(s-1)/ν 2 ) 1/3 )) dimensionless particle size ν kinematic viscosity (m 2 /s) Fr (=V/(g(s-1)/d) 0.5 ) densimetric Froude number ζi, ζi* slack variables g gravitational acceleration ρ water density (kg/m 3 ) g(x) membership function ρs sediment density (kg/m 3 ) H neural network output matrix φ a nonlinear function K(xi-xi * ) kernel function Subscripts N ~ number of hidden neurons s sediment R hydraulic radius (m) 1. INTRODUCTION 1 One of the most important issues in open channel design is the economic and optimized planning of it. 1 *Corresponding Author’s Email: bonakdari@yahoo.com (H.Bonakdari) Due to the through path of flow before reaching the channel, the inflow may erode and suspend sediments which are then transported with the flow into the open channel. If the flow velocity for a given channel slope (limiting velocity) is insufficient to transport the sediment in the flow, the sediment will be deposited within the channel. In the case of fine sediment, the